THE COMPARISON OF THE ANTICIPATED AGGREGATE COSTS WITH THE ANTICIPATED AGGREGATE BENEFITS OF AN ACTIVITY IS THE SIMPLEST DESCRIPTION OF BENEFIT-COST ANALYSIS. GENERAL LIMITATIONS OF THE BENEFIT-COST ANALYSIS ARE (1) INPUTS AND OUTPUTS ARE EMPHASIZED WITHOUT EXPLICIT CONSIDERATION OF THE PROCESS LINKING THEM, (2) DIFFERENTIAL EFFECTS OF PROGRAMS ON CLIENT SUBGROUPS MAY BE IGNORED, AND (3) PROGRAM ADMINISTRATORS, WHO MAY HAVE LITTLE TO GAIN FROM POSITIVE FINDINGS BUT MUCH TO LOSE FROM NEGATIVE FINDINGS, FEEL THREATENED. A SPECIFIC LIMITATION OF USING BENEFIT-COST ANALYSIS IN CORRECTIONAL PROGRAMMING IS THE DIFFICULTY OF ATTACHING MONETARY VALUE TO THE BENEFITS TO BE DERIVED FROM MANY CORRECTIONAL PROGRAMS. MANY OF THE COSTS AND MOST OF THE BENEFITS MUST BE VALUED THROUGH THE USE OF 'SHADOW PRICES.' IF THERE IS NO CONSENSUS ON HOW PROGRAM EFFECTS ARE TO BE VALUED ECONOMICALLY, THEY CANNOT BE COMPARED IN TERMS OF BENEFIT-COST RATIOS. EVEN THOUGH SOME INPUTS AND OUTPUTS RELATED TO A GIVEN PROGRAM MAY NOT BE MONETARILY QUANTIFIABLE, THE BENEFIT-COST ANALYSIS DOES DICTATE THE IMPORTANCE OF IDENTIFYING EVERY ELEMENT OF INPUT AND EVERY ANTICIPATED BENEFIT, WITH AN EFFORT MADE TO DETERMINE AN APPROPRIATE AND QUANTIFIABLE MEASURE FOR EACH. SHOULD THE BENEFIT-COST ANALYSIS BE USED IN COMPARING ALTERNATIVES, THE PROGRAMS MUST BE SUCH THAT UNMEASURABLE EFFECTS WILL BE SIMILAR. ALSO, THE PROGRAMS CONSIDERED SHOULD BE EVALUATED THROUGH AN EXPERIMENTAL OR STRONG QUASI-EXPERIMENTAL DESIGN. AN EXAMPLE OF HOW THE BENEFIT-COST APPROACH CAN BE USED IN A CORRECTIONS EVALUATION IS PROVIDED FOR A SUPPORTED WORK EXPERIMENT FOR EX-OFFENDERS IN NEW YORK. SUGGESTIONS ARE OFFERED FOR FURTHER ECONOMIC RESEARCH THAT CAN AID IN INCREASING THE EFFICIENCY OF THE CORRECTIONAL SYSTEM. REFERENCES AND A BIBLIOGRAPHY ARE PROVIDED. (RCB)